Villani’s Report: Defence at the Age of AI
29 mars 2018
After more than six months of work, hearings and contributions of all kinds, the report of the parliamentary mission led by Cédric Villani has just been submitted to the government. Among the fields covered, Defence occupies an important place which underlines the evolution of the French strategy.
At the end of this report (“Giving meaning to artificial intelligence, for a national and European strategy”), defence and security issues are the subject of a chapter that sets the framework for future transformations to be undertaken by French Defense to turn the corner of artificial intelligence (AI). The authors predict that the mastery of AI technologies will become a necessity for the state, to "ensure security missions, maintain the ascendancy against our potential adversaries" and "maintain our position in relation to allies ".
AI Civil-Military Complex
More importantly, the report highlights the need for close collaboration between the public and the private sector, in order to bring out "quality ecosystems"; in other words, building synergies around civil and military innovation in IA: recent announcements by the Minister of the Armed Forces, Florence Parly, around the creation of a "Defense Lab", are similar. To achieve this result, the different entities from the public and private sectors will have to exchange their data, the raw material of Machine Learning. Obviously, this will lead to technical difficulties, in addition to the legitimate security and confidentiality issues that could be raised by the sharing of certain sensitive data. Nevertheless, the Villani Mission and the Minister of the Armed Forces seem intent on decompartmentalizing AI research and innovation.
The stated goal is to build a civil-military complex of technological innovation, focused on digital technology and more specifically on artificial intelligence. In contrast to previous decades when the results of military public research were pouring into the civil sphere, the civilian sector is now leader in research and innovation. The authors of the report have got the measure of this major shift, that’s why they invite to implement the best adapted tools to facilitate the transfer of technologies from one sphere to another.
A Stake of Sovereignty
The combination of research efforts between the civilian sector and the military field is all the more important because it represents a major issue of sovereignty. GAFAM (Google, Amazon, Facebook, Apple, Microsoft, but also IBM) are the main players in artificial intelligence, and their complete supremacy places most states and people around the world under their control. In France, over a month, nearly 80% of visits to the 25 most popular websites are captured by major American platforms (see study of the Castex Chair of Cyberstrategy); these are all data collected and valued by American firms that the State doesn’t control. The case of the American start-up Palantir, linked to the CIA, specializing in the analysis of big data and which signed a contract of 10 million euros with the DGSI, is another indicator of the American domination in IA on the French Defense and Security sector.
To face up to the technological power of GAFAM, which also has close links with the Pentagon, the Villani’s report suggests strengthening "our ability to cross massive amounts of data", because "the amount of data available and the quality of their annotations are key to advancing research on AI applications". In order to do this, the option of the “pooling platform” is then privileged. It would enable military, scientific and economic actors to pursue joint programs of research, experimentation and innovation with "access to operational data within an administration-controlled framework". With this in mind, Florence Parly announced the creation of a defence innovation agency "largely open to the civilian sphere".
The Chinese and American Paradigms
The recommendations of the Villani report for the development of artificial intelligence in general, and military AI in particular, are almost perfectly in line with the strategies deployed by the two giants of the sector: The United States and China. The report praises the Defense Advanced Research Projects Agency (DARPA), to praise its ability to appoint recognized experts in their field to lead research and development programs, and for its risk culture and its ability to launch bold projects.
Reading this report, we can even be surprised to find a rhetoric very similar to that used in the US by Eric Schmidt, former Google’s CEO and, since 2016, chairman of the DoD Innovation Advisory Board. On several occasions, Eric Schmidt pointed out to the bureaucratic ponderousness and archaic DoD culture that hinder his ability to innovate, particularly in the field of AI. Considering the rapid progress of China, the American businessman has been urging his administration for months to open up to innovative firms in the private sector, failing which, in his view, the technological leadership of the United States would threaten to fail. In a more refined language, the authors of the report seem to draw the same conclusions: "For historical reasons, the current operational systems are mostly not designed to implement this research and innovation open to the ecosystem.”
More interestingly, the report suggests development paths very similar, in some respects, to the AI strategy deployed by China. At least since the launch of its "Next Generation Artificial Intelligence Development Plan" in July 2017, Beijing has made "civil-military fusion" the major focus of its national AI strategy. Technological firms (Baidu, Alibaba, Tencent, Xiaomi) and innovative startups (iFlytek, Horizon Robotics, SenseTime) are now heavily relied on by the People's Liberation Army, which intends to take advantage of the private sector's most exciting advances in AI to extract military applications.
The authors of the report are therefore very clearly inspired by models that have proved their worth. It is no coincidence that the United States and China are today the undisputed leaders of artificial intelligence, and that their technological advance in the civilian field allows them to develop military AI applications. With this report, France shows that it has certainly taken a decisive step. But, as its authors rightly recall, its competitiveness in the civil and military fields will undoubtedly depend on its ability to gather around this initial impulse a real European synergy. Without European convergence, French AI will remain in its infancy.